A Vehicle Tracking System Using Greedy Forwarding Algorithms for Public Transportation in Urban Arterial

A vehicle tracking system assists public transportation users in their movements by providing real-time information on the locations of vehicles in transit. Public transportation in parts of developing nations, especially Nigeria is ineffective. The system is chaotic and frustrating, especially at peak traffic periods. In a bid to address the problem, a vehicle tracking system was developed as a component of an Advanced Public Transportation System to improve commuting in an urban arterial. The developed system is based on wireless technologies of the Global Positioning System (GPS) and Global System for Mobile Communication module. It records and displays real-time vehicle location using a GPS-based greedy forwarding algorithm, computes route distance information using distance-time based algorithm and radar range sensor (RRS). A pseudo-range mathematical model using the Haversine formula was adopted in determining the accurate position of an object during signal transmission from GPS satellites to the receiver message module. The minimum inversion matrix method was used for the GPS-based geometric dilution of precision (GDoP) selection of satellite approximation and distance. Atmega328P controller chip was used as the logical control unit for processing activities in the tracking system and programming in Arduino IDE using C-language. The system was deployed to a university transportation system in Nigeria: a journey to and from the Bosso and Gidan Kwano Campuses route in the Federal University of Technology, Minna. The vehicle tracking system was tested with 11 tracked satellite and minor dilution error (PDoP error = 1.9, HDoP=0.9, and VDOP=1.7) was recorded. The system is efficient and accurate in distance and time information display with a minor delay. The system would enhance fleet management schemes for urban arterial and can be adopted universally.


I. INTRODUCTION
The public transport system (PTS) remains the easiest and cheapest means of moving passengers or groups of people from one point to another in vehicles. Its transformation is indispensable and it was reported [1] that about 20% of the gross domestic product of Europe equating to billions of euros and millions of jobs is generated through the transport The associate editor coordinating the review of this manuscript and approving it for publication was Venkata Ratnam Devanaboyina .
sector. Also, the Census Bureau Statistics of 2019 in the United States of America (USA) claimed that the average citizen spends about 25.9 minutes per day on traveling to work and four hours per week in transit through the PTS, and over 15.9% of his budget on transportation [2]. In line with the trend in PTS, Sub-Saharan African countries are promoting the system. For example, the Trans-Africa project is focused on the provision of sustainable and efficient transport systems in Africa cities for improving the production process and growth, as well as reducing congestion and traffics on the roads [3]. Effective PTS as well as information and communication technologies (ICT) are essential to national economic growth and social development [1]. PTS are typically managed on a schedule and functioned on time-honored routes [4]. It may be operated on a trip charge fee as found in the city buses, tram (a rail vehicle along public urban streets that runs on tramway tracks), trolleybuses, ferries, mass rapid transit (MRT), and metro [5], [6]. Several technologies have emerged over the past decade in the form of real-time vehicle tracking systems that utilize remote wireless expertise techniques like GPS, GPRS, and other navigation systems for determining vehicle location and enhanced fleet management. These wireless technologies operate through satellite and ground-based stations [7], [8].
Software tools have been developed to improve fleet productivity [9]. Fleet tracking device offers numerous advantages such as high visibility of geographic location, monitoring the vehicle speed, and other related activities [10]. It also assists in vehicle tours planning, optimizing, and improves customer services (passengers) by keeping passengers informed about the journey distance and arrival time. Other benefits include business workflow scheduling and service delivery period. A tracking system is capable of rendering virtual space to a human through the observer's coordinates [11]. Other functions of the vehicle tracking system in fleet management support include displaying vehicle speeding information reports, harsh braking, driver's acceleration, and the route which can be downloaded to a computer and used for analysis in the future [12].
The understanding of environmental correlations associated with vehicle route selection and scheduling during transportation was discussed with a useful tip [13]. By relating the features of the route linkages between the possible source path to the shortest destination way. An improved vehicular routing protocol that considers the street/route density and direction called GyTAR (Greedy Traffic-Aware Routing) was discussed. It utilizes GPS to determine the vehicular speed and position on a digital map [14]. The connectivity of the vehicular pathway between the source and the destination track was model and discussed as illustrated in Fig. 1. This is to measure the average delay of the vehicle on the transit using a Genetic Algorithm (GA) as an optimization routing trail technique for the vehicular area network [15], [16].
The investigation about the choice of early morning commuter route behavior using Global Positioning Systems (GPS) and Multi-Day Travel Data (M-DTD) approaches was reported [17]. This considered the factors that influence early morning commuters traveling choice and route switching based on objective observations of travel behavior during a multi-day period in the real-world.
The socioeconomic characteristics is another influence in fleet management that comparing the transportation systems in Maputo and Nairobi [3]. The demand and supply of bus services in each city were influenced by the mode of transportation trips. Two logistic regression models were studied and analyzed. It was discovered that age decreases  the likelihood of choosing buses and residence location in Maputo. Other factors affecting bus shuttle customized in both cities include employment, income, vehicle ownership, and residential location.
Some of the universities in Nigeria operate campus systems with students on off-campus residence arrangements. This results in a transportation problem for movement into and between campuses. Therefore, Federal University of Technology, Minna (Nigeria) operates on two campuses (Gidan Kwano and Bosso), a distance of 28 km apart. The Bosso Campus is cited within the city where students find suitable accommodation for off-campus residency. About half of the students in Gidan-Kwano are resident in the Bosso area. The implication is that there are captive commuters seeking transportation services from Gidan-Kwano and departs at the respective terminal when filled without on schedule. But only registered vehicles are allowed to operate as public transport into the campuses. Fig. 2 illustrates a typical passenger frequency into the two terminals, which is a time-based schedule and found to be uneconomical because traffic is one-way at certain times. The solution to the identified technical challenges in this public transportation system would involve an economic schedule system and dissemination of information to commuters remotely. This paper addresses the component of the dissemination of information to commuters and operators remotely.
A geographical map location-based vehicle tracking system was developed and tested on the public transportation system between the Gidan-Kwano campus and the Bosso campus. It is a component of the fleet management system to improve the efficiency and effectiveness of transportation as the first stage in a confrontation of this universal challenge. The objectives include: (i) providing real-time tracking of the bus shuttles on inter-campus metropolis routes, and (ii) displaying vehicle location and arrival time at terminals. Different algorithms and mathematical modeling were adopted and considered for the geolocation coordinate tracking, distance calculations, and display of remote geo-position information on the dashboard and google map respectively. In the future objectives, the traffic route (noise) will be studied and analyzed using an optimized technique for the efficient time table development in scheduling. A digitized Google map that illustrates the details of the geographical location of the bus parks at both campuses with route analysis was presented in Fig. 3.
In this research, a minimum inversion matrix geometric dilution of precision (minGDoP) method was adopted in the GPS satellite selection for accuracy and precision. More emphasis is on the development and implementation of the schedule vehicle tracking system for public transportation on urban arterial using different novelty algorithms, mathematical computations, and modeling of distance detection width estimation as contains in section III. This includes distance-time based parameter tracking using pseudo-range and Haversine formula. Modeling of the distance detection width calculation was done using a GPS-based greedy forwarding algorithm for the estimation and in the selection of the best possible response at every distance covered with time relation. Section IV discussed details of system design and implementation for both hardware and software coding with performance evaluation. This section includes a novel algorithm for remote transmission and distribution of vehicle data (messages) called wheel wagon data dissemination protocol. Section V presents detailed results and discussion, and section VI concludes the research work with a recommendation and future works.

II. OVERVIEW OF RELATED WORKS
The vehicle tracking system is a technology used in monitoring, safeguarding, and providing directions for both private and public commuter users. It enables the owner to virtually keep an eye on the vehicle arrival through the route and also aids the public commuter users to manage their time resulting in cost-efficiency. It can determine the exact position of space object which includes longitude, latitude, and height [19]. Data from a GPS-based tracking device can be viewed on electronic maps, google maps, or mobile applications connected to the internet. Several techniques can be used in tracking vehicles on the arterial route and include indicators or 'lag time'. For example, using a bar code or choke point or gate for vehicle data collection when passed a point. The bar-code systems require scanning of the personal items for automatic identification using RFID autoid. A real-time tracker like Global Positioning Systems (GPS) operation is another type which depends on how often the data is refreshed or updated for prompt display. The most common world tracking system consists of discrete hardware and software systems for different applications [20].
GPS-based or assisted GPS (AGPS) sensors technology for the vehicle tracking system can be classified into three parts as illustrated in Fig. 4.
Both active tracking technology (energetic) and passive tracking technology (reflexive) use GPS or assisted GPS sensors like radar sensors for execution. The energetic tracking systems utilize wireless connection modules that include Global System for Mobile communication (GSM) and General Packet Radio Service (GPRS) for data transmission over the internet. The components besides the central server (database) receive, process, and store the generated information [21]. But, the integration of both components (GPS and GSM) into a single module was adjudged the best in terms of cost and configuration [22]. This active tracking technology can be classified into three types, which include Automatic-Vehicle Location System (AVLS), Assisted-Global Positioning System (AGPS), and Radio Frequency Identification (RFID) technology [23]- [25].
The reflexive (passive) tracking devices focus majorly on downloading, recording, and storing tracking information using a GPS module for future view [26]. This type of tracking technology is embedded with onboard memory but does not report or function in real-time. The GPS-based tracking devices are widely adopted in the vehicle tracking technology, security, energy-harvesting tracker (solar-energy tracker), and geolocation coordinate tracing [27], [28]. The manual type is locally designed majorly for solar panel trackers, surveillance, and so on. An automobile tracking system was developed [29], [30] to secure oil and gas distribution using the telematics approach and blockchain technology. The tracking system addressed the high-level susceptibility of illegal diversion of automobile crude oil conveyance along the route. A telematics approach using GPS/GPRS and GSM module was used in tracking the geo-location, speed, and volume of products conveyed by the automobile system. The performance of the GPS systembased satellite tracking was evaluated based on the success rate, sensitivity, and location accuracy.
In the study [31], a mobile phone-based vehicle positioning and tracking system were developed for an urban traffic state estimation with an emphasis on the mobile positioning and navigation of the vehicular environment. The Kalmanfilter based hybrid technique was adopted in tracking and positioning the mobile phones traveling onboard vehicles. The simulation processes of the universal mobile telecommunications system (UTMS) was carried out in the MATLAB environment using combined standard methods of observed time difference of arrival (OTDOA) and assisted global positioning system (A-GPS) location estimate as to the state vector level. The simulation result statistically demonstrated that the hybrid technique of the Kalman-filter method shows better performance in position determination and velocity estimation.
A simulation analysis of geographic location and distance routing on the vehicle area network (VANET) was carried out [32]- [34] using the wireless technology approach (IEEE 802.11p). The work addressed the issue of routing in VANET due to road constraints like over-speeding of vehicles and road barriers. The system performance evaluation was based on the routing overhead, throughput, packet loss, and delay using network simulation-2 (NS-2). The inversion matrix technique is commonly used in GPSbased GDoP selection of satellite to achieve an optimal approximation subset [35], but the computational complexity is highly intensive to be practical and it is timeconsuming. A stochastic global search algorithm like the Genetic Algorithm (GA), Artificial Neural Network (ANN) has been used for the optimization and fast selection of GPS-based GDoP approximation [36]. Although the ANN approach gives better performance, it requires a long time in training and ends up with difficulty in the architecture determination. The performance evaluation of adopting GA in GPS-based GDoP satellite selection is fair when compared with the inversion matrix or Neural Network (NN) approach. The GA application offers a reduced cost of calculation, fast and precise in GPS-GDoP classification, and approximation [37]. Other related works with modeling and simulation of vehicular area network tracking are presented in Table 1.

III. MATERIALS AND METHODS
The development of an advanced public vehicle tracking system for captive commuters on urban arterials using geographical map location and GPS for tracking involves hardware component integration with software design coding. A vehicle tracking device was embedded in each vehicle and a google map location software was designed for monitoring vehicle positions, data collection in the field, and delivering information to visual display stations. In the visual display station, a Raspberry Pi ARM Cortex A53 processor with a universal modem was embedded and utilized for the ease of inter-communication between the base station and remote area. The Atmega 328P was used as the central controller for the in-vehicle tracking device and the programming was achieved in the Arduino Integrated Development Environment (AIDE) using C-language. The advanced public transportation statistics system for data collection, data fusion, and information dissemination are illustrated in Fig. 5. The system operation flowchart is presented in Fig. 6.
The real-time data acquisition and vehicle location algorithm for a vehicle in transit areas contained in Table 2. The bus shuttle operates on a demand-supply basis. The minimum safe headway (that is, a distance measured with time) was calculated by the decelerating performance using Anderson techniques as in ''(1)''. The total time taken for the vehicle and headway is calculated as in ''(2)''.
where ℵ min−safe is the minimum safe headway in seconds, R time is the reaction time (the maximum time it takes for the following vehicle to detect a malfunction in the leader and to fully applied emergency brakes), k is an arbitrary safety factor which is ≥1, V is the speed of the vehicles in (m/s), α f is the minimum deceleration braking of the follower (m/s 2 ), β l is the maximum deceleration braking of the leader in (m/s 2 ), L  is the vehicle length (m) and T total is the time for a vehicle (s) and headway to pass a point.

A. PSEUDO-RANGE METHOD
The pseudo-range method was used to determine the accurate position of a vehicle during signal transmission from GPS satellites to the receiver message module as in ''(3)''.
where, α p , α q , α r are receiver positions for the components of (p, q, and r), v is the speed of light (velocity), T x is the time difference between the receiver and satellite. P 1 , Q 1 , R 1 are known as the three components of satellite positions, and ρ i is the accurate distance of ith satellite from the receiver in ideal situation. The pseudo-range computation determines the difference between satellite and GPS receiver [38]. The expression in ''(3)'' can be re-written using Taylor's series as expressed in ''(4)'', ''(5)'' and ''(6)'', respectively. If the initial coordinates of the receiver are known to be (p 0 , q 0 , r 0 ), the actual receiver coordinates are (p 0 , q 0 , r 0 ) at epoch time 1 (t 1 ). The summary of the developed formula for all the satellite observation can  be given as in ''(7)''.
where, p j , q j , r j is the coordinate of the jth satellite, ρ j i indicate the distance between the jth satellite and the pseudorange between satellite, and receiver at epoch time t 1 The distance-time-based parameters (DTBP) approach was used for the detection of geo-location tracking of vehicle position [39]. The Haversine formula [40], [41] was used for the computation in determining coordinate (latitude and longitude) distance of a great circle as illustrated in Fig. 7. This is because the time-distance based parameters offer the advantage of accuracy and flexibility in the server update. The Haversine (ξ ) formula is used in the algorithm for calculating the reference points of the geo-location coordinates (longitude and latitude) to reduce rounding errors that may be generated before communicating with remote locations (refer to the expression in ''(8)'').  The distance d from ''(8)'' is solved using the inverse sine function as in ''(9)'' and ''(10)''.

B. MODELING OF DISTANCE DETECTION WIDTH
The radar sensor was used to measure the distance covered. This sensor sends a command to the digital signal processing board (controller) through a serial port about the velocity of target detection and vehicle positions. The mathematical modeling [42], [43] for the distance detection ranges estimation of a vehicle in motion using the radar sensor is presented in Fig. 8. The motion characteristic for determining the vehicle states which is either in dynamic or stationary is given in '' (11)'', and can be computed using discrete Kalman filters expression in [44] as in ''(12)''.
VOLUME 8, 2020  where x and y represent longitude direction and lateral positions of a fixed coordinate body respectively. κ is curvature, ϕ is the angle that exists between the road lane and longitude axis of the vehicle sensor, v is the vehicle speed, µ is the yaw rate, is the range rate, subscript i is the ith radar track.
x(κ + 1 | κ) = Ax( κ| κ), The updated equations for the measurement along the route or lane can be expressed as in ''(13)''. Then, since a constant speed or velocity is considered with a vehicle in motion,   longitude axis and lateral velocity or speed, and T is sampling time.

C. GPS GREEDY FORWARDING ALGORITHM
In the vehicle tracking, Greedy Forwarding Algorithm (GFA) was adopted in selecting the best possible response at every distance covered per kilometer in relation to time, the subsequent steps were calculated until the vehicle reaches its destination as illustrated in the decision tree of Fig. 9. The route colored in blue was identified as the optimum and best route to the destination. The GPS-GFA is presented in Table 3. This algorithm assists in achieving the globally optimum path selection [45]. The Haversine formula was used for the accurate computation of the distance between the two points (latitude and longitude) of the great circle as in ''(10)''.

IV. SYSTEM DESIGN AND IMPLEMENTATION A. HARDWARE COMPONENTS INTEGRATION
The combination of components used in the design of the Public Transportation Vehicular Tracking System (PTVTS) includes the Atmega 328P controller, Sim 900 GSM module, Ublox NEO6MV2 GPS module, Button, connecting wires, and power supply. Other standalone system involved in the PTVTS architecture includes smart LED TV, android mobile phone, Huawei universal MiFi, Raspberry Pi (ARM cortex A53) processor and GPS cellular network as VOLUME 8, 2020 illustrated in Fig. 10. The block diagram of PTVTS is illustrated in Fig. 11. The flowchart in Fig. 12 illustrates the logical principle of the tracking device operation.

1) COMMUNICATION TRACKING SYSTEM MODULE
The in-vehicle tracking device transmits the geo-location information to the remote base-station through the integrated Global Positioning System (GPS) and Global System for Mobile communication (GSM) module with the system controller (Atmega 328P chip). The GPS module consists of three components that are responsible for navigation in tracking such as the airspace section, control section, and use section [46], [47]. A GSM component (SIM 900) is an embedded memory chip with available space for the subscriber identification module (SIM) card. This module was used to establish communication (that is, the transmission of data rate) between a GPRS and computing system. The essential pins connection and configuration of this module with ATmega328P for message forwarding and receiving includes +5 power (PWR), ground (GND), Receiver (RX), transmitter (TX), and reset (RST) button. The remote transmission and distribution of vehicle data (messages) were achieved through the adoption of data dissemination protocol using Wagon Wheel (WW) approaches for the segmentation of message transmission range [19], [48]. The proposed protocol moderates the message broadcast storm communicated from different vehicles on the route and controls some redundant transmissions as shown in Fig. 13. The WW data dissemination algorithm guarantees the distribution of vehicle messages (data) across the network partitions and periodically connects through the available vehicle service on the route as presented in Table 4.
The vehicle position ( ), distance (∂ st ), and direction (γ ) are the parameters coding in routing information to the target remote display board. The cumulative value η for each or individual vehicle message received R i on the urban arterial routes is given in '' (16)''.

2) THE CONTROLLER MODULE
The Atmega 328P microcontroller was used as the logical control unit that processes all the activities in the tracking system. This controller was integrated on the Arduino board with 80mA, 5volt power when no component is connected.
The microcontroller board can be powered either from the DC power jack (7-12V), the USB connector (5V), or the Vin pin of the board (7-12V). The maximum current draw is 50mA from the supply voltage of 3.3-5V, while the clock speed is 16 MHz so it can perform tasks faster.

3) THE DISPLAY SYSTEM UNIT
The smart light-emitting diode television (LED-TV) was adopted as a dashboard module for displaying information. This system was subjected to an engineering re-design with the integration of some components. The process involved integrating a microcomputer system (Raspberry Pi ARM Cortex A53 processor) into the smart television for logical control and receiving remote geolocation information from the GPS/GSM module. The collection and display of remote information were achieved through a High-Definition Multimedia Interface (HDMI) connection enable and universal mobile wireless communication device, which was connected through the universal serial bus (USB) port to provide a stable/strong network The precision was determined to be the difference between two coordinates distance measured for a fixed location which is 1.24 meters for the received data. (received signal strength RSS) for the consistent functioning of the microcomputer system. The Raspberry Pi ARM Cortex A53 processor is a 32-bit quad-core processor that operates on the frequency of 900 MHZ. It supports Android OS, Linux, OpenBSD, RISC OS, Windows 10 ARM64, and Windows 10 IoT core. It utilizes the CPU of 1.5 GHz 64/32-bit quad-cores, a micro-SDHC slot with 4GB LPDDR4-3200 RAM [49]. The microcomputer is a system-on-chip (SoC) that uses Broadcom BCM2711 and operates on a full power delivery to USB devices of 5V/3A as illustrated in Fig. 14 with ports connection pins.

B. SOFTWARE SYSTEM DEVELOPMENT
The software design coding for hardware components, C-language was used to program the microcontroller and other peripherals including the GPS module and GSM in Arduino IDE. Google maps geolocation tracking was developed using Python programming language. The user interface designs for the system utilized Microsoft Visio, while the implementation was programmed using HTML, CSS, JavaScript, and Python. The HTML was used to provide the visual elements on the webpages, CSS was used to style the content to provide good aesthetics, JavaScript for HTML manipulation, and to provide meaningful content for the webserver. Python was adopted to manipulate and process the real data (from the database) in the web hosting sites services (Website). The flowchart for geo-location information display is depicted in Fig. 15. Also, virtual network computing (VNC) viewer software was used as virtual control of the dashboard message display. A VNC is a graphical inter-relaying screen (desktop-sharing screen DSC) direction on the network, which employs a Remote Frame Buffer Protocol (RFBP) to control another computer in a network remotely [51].

C. PERFORMANCE EVALUATION
The accuracy and precision of the system performances were carried out using the Ublox Neo 7M GPS receiver software. The vehicular position tracking accuracy of the system was defined in terms of GDoP [52], which is the ratio of the square root of MSE to standard deviation as expressed in '' (17)'', ''(18)'' and '' (19) The GDoP is a non-linear function that can be linearized with the Taylor series mathematical model [53] as in '' (20)'' and given in a matrix form ( ) as in '' (21)''. We assumed that the measurement error of GDoP has an independent Gaussian distribution, so the least-squares error solution for the matrix is given as in '' (22)''.
For the position accuracy, it can be proven in terms of covariance ( ) of x as in ''(23)'' to ''(25)'': Therefore, it is rational to assume that all the errors in the pseudo-range measurements are motionless of the random processes for a short period. They are identical and independent distributed with a variance of σ 2 and is expressed as in '' (26)''.  The amplification of the equivalent error range measurements of the receiver position is calculated using GDoP as given in '' (27)''.
The matrix analysis and applied theory of linear algebra in [54] considered that, if the λ i is recognized as the eigenvalues of a converse matrix , then it is expected that λ −1 i it can represent eigenvalues of the inverse matrix of −1 . Therefore, the matrix measurement can be expressed as = ( T ψ) −1 which is always reversible and be solved using λ i ( T ψ). This process can result in a significant reduction of inverse matrix calculation using QR decomposition with Gram-Schmidt [55] for geometric dilution of precision (GDoP) as given in '' (29)'' to ''(30)''. Therefore, the visible selection process of the satellite in the present view for the best positioning accuracy corresponding to the row selection of the matrix form ψ observability will result in minimum geometric dilution of precision (GDoP) as in ''(31)''. S − {s i }, 1 ≤ i ≤ k, k ∈ (4, 5, 6, 7).
where is the trace function of the matrix, c is the constant speed of light, s i is the number of identifying satellite that is currently view. The Position Dilution of Precision (PDoP) is given as where Vertical Dilution of Precision, VDOP = σ 2 z , and Horizontal Dilution of Precision HDOP = σ 2 x + σ 2 y . For the local horizontal and vertical plane corresponding, x, y, and z are coded for the down coordinate system (North, East), and up coordinate system (East, North).

V. RESULTS AND DISCUSSION
The developed Google Map API using python language display information about the vehicles on transit (such as vehicle ID, origin, destination, and arrival time) as shown in Fig. 16.  The simulation result carried out in U-Blox GPS software shows the accuracy and precision of the geo-location coordinate information achieved. The circular area is 15 meters apart, and the satellite selection of the geo-location accuracy varied between 7.5 meters and 10.5 meters of the circular radius as illustrated in Fig. 17. The precision was determined to be the difference between two coordinates distance measured for a fixed location which is 1.24 meters for the received data. The sensitivity to satellite and success rate is presented in Fig 18. The vehicle tracking system was tested with different sample sizes of GPS received information and simulated in the Google View Software. The GDoP errors such as (PDoP), (HDoP), and (VDoP) for the different sample sizes as illustrated in Fig. 19. In Fig. 19(a), 11 satellites were viewed but only 5 satellites could be tracked as a result of poor network, resulting in PDoP error = 4.7, HDoP error = 3.4, VDoP = 3.1. The result shows a fair navigation in-route precision. In Fig. 19(b), 13 satellites in view but 11 was tracked resulting VOLUME 8, 2020  in minor dilution error. PDoP error = 1.9, HDoP = 0.9, and VDOP = 1.7. This result is excellent, and the positional measurement is considered accurate and precise based on the principle of the dilution error range. In Fig. 19(c), 13 satellites in view, 11 tracked with minor dilution error. PDoP error = 2.0, HDoP = 0.9, and VDoP = 1.7. This result is also excellent, and the positional measurement is accurate and precise based on the principle of the dilution error range.
The geographical location tracking based on the latitude, longitude, and altitude is presented in Table 5, with the average prediction in Table 6. The simulation was carried out in Google View software to evaluate the system performance on the coverage area of the satellite precision and accuracy. The graphical results are illustrated as in Fig. 20 and 21. The HDoP error from the simulation analysis for different samples is presented in Fig. 22.  The geo-location and positional information were sent remotely to the server database and displayed on the dashboard with the aid of a Raspberry Pi ARM Cortex A53 processor integrated with smart LED television using Hyper-text Transfer Protocol (HTTP). The vehicle tracking position information displayed is presented in Fig. 23. Table 7 contains information received during system testing which includes time, date, speed, longitude, latitude, and distance. The implementation cost of this proposed vehicle tracking system for public transportation on urban arterial can be widely adopted and scaled up to a minimal amount of $2,570.00. The comparison analysis result of satellite selection sensitivity using different techniques for the geolocation vehicular tracking system is presented in Table 8.

VI. CONCLUSION
The vehicle tracking system for public transportation on urban arterial was developed, implemented, and demonstrated using novelty algorithms. The GPS-GFA receiver was used to collect data on remote vehicle state (geo-location, distance, and time to arrival). GSM module was used to transmit information to remote server base station using wheel wagon data dissemination protocols (WWDDP). The general packet radio service and Google Map software were adopted for geo-location tracking and identification. Both the time and distance to arrival achieved were found to be accurate with low latency. The minimum inversion matrix GDoP method for GPS satellite selection approach proof efficient with an excellent performance recorded in the measurement of accuracy and precision. The vehicle tracking system was tested with 13 satellites in view but 11 was tracked with minor dilution error (PDoP error is 1.9, HDoP is 0.9, and VDOP is 1.7 respectively). Therefore, the developed vehicle tracking system for public transport on the urban arterial route can be scaled up for city-wide application with minimal cost of $2,570.00 per terminal. This system would help to reduce unnecessary anxiety experienced while waiting for vehicles at bus terminals/stops without knowledge of the vehicles' locations and arrival times. It would help to guard against diversion of fleet or vehicle to unauthorized routes, enhance fleet management, and reduce environmental pollution. Future work will be focusing on the development of an intelligent transport fare billing system, analysis of traffic (noise) on urban arterial routes, and scheduling of bus shuttle timetable.